Numerical models of the interaction lithosphere/mantle Klaus Regenauer-Lieb, Dave Yuen*, Francesca Funiciello, Gabriele Morra Institute of Geophysics, ETH Hönggerberg,
8093 Zürich, *Supercomputer Institute, Univ. of
Minnesota, Minneapolis,
[email protected] , phone +41 633 2058, fax +41 1 633 1065
Abstract Recent advances in computational geodynamics give directions to solving the coupled lithosphere deformation/convecting mantle problem. At the heart of this lies the "subduction" theme, which deals with cold solid lithosphere plunging into a convecting fluid like substance. Lithosphere dynamicists tend to assume that to first order subduction is governed by the rheology of the lithosphere itself responding to the negative buoyancy forces arising from it. Convection dynamicists solve the problem more comprehensively by acknowledging plasticity within the lithosphere as well as flow in the deeper mantle. Unfortunately, the solid mechanical database is neglected in those more comprehensive approaches. Here, the problem is solved in a somewhat mixed approach, i.e. a solution is presented in which a coupled solid-mechanical /fluid-dynamic lithosphere rides on a passive mantle (without active convection). The setup is used to investigate self-consistently the problems of subduction initiation, the subsequent "freely" falling slab and the interaction
with
the
660km
discontinuity.
The
approach
gives
new
insights
into
coupling/decoupling of deformation within the lithosphere as a function of wet/dry rheology during subduction initiation, the "elasticity" of the slab, its coherency and subduction angle as well as the penetration or non-penetration problem of slabs into the 660km discontinuity.
Introduction Our angle of attack to the coupled solid/fluid mechanical problem is to present a numerical approach that is suited to model within one and the same finite element calculation the response of the lithosphere to any kind of plate tectonic or other loading. The approach is also numerically inexpensive and resolves all necessary degree of complexity from regional (several 100m) to plate tectonic scale (several thousand km). Furthermore, our approach accounts for a viscous time scale of the underlying "asthenosphere" missing in previous solid mechanical approaches. Finally, it is free of boundary problems posed by finite size of numerical boxes when mantle convection is considered explicitly. The method is simple: we replace the convecting mantle by one dimensional elements and thereby bypass the volume flux problem. Instead of a 2-D box we use a family of nodal x-z dashpots that respond by a velocity dependent force.
Figure 1. Numerical setup of the lithosphere resting on x-z dashpots representing the asthenosphere. The lithosphere is modeled down to the 1300 K isotherm, i.e. its fluid like part is included in the model.
Subduction dynamics The approach has been applied to the entire history of subduction mechanics from initiation to slab breakoff. Here, we show a that it can be used for resolving the long standing problem of weak zone creation for subduction initiation. A separate contribution will deal with the dynamics of slabs after subduction initiation [1].
Feedback and weak zone formation Feedback is the key to investigate the problem of weak zone formation self-consistently [2] [3]. Therefore we must use the most sophisticated data for lithospheric rheology considering thermo-elastic and thermal-mechanical feedback of a visco-elasto-plastic solid/fluid in a high resolution (600m) calculation. We construct a hypothetical Atlantic type continental margin where the lithosphere is loaded by a steady flux of sediments (maximum thickness of 15km at 100 Myrs loading time). Initial thermal conditions are defined with reference to the analytical cooling half space model. No geometrical/chemical seed for shear zone nucleation has been assumed. Water fugacity is varied as a free parameter and the rheological response of the lithosphere is monitored. Our hypothetical passive continental predicts nucleation of a weak zone for subduction only if the water fugacity is raised above that of a nominally "dry" lithosphere [4]. The dry lithosphere case predicts a stiff elastic core inside the lithosphere that communicates surface loads (sediments, volcanic plateaus etc.) through to the fluid dynamic layer. The bottom of the oceanic lithosphere finally drops off as a Rayleigh-Taylor instability having its own deformation time scale. The most striking result of our dual feedback model is the generation of extremely narrow shear zones in both wet and dry cases (Fig. 2). Several splays of dip slip faults converge on a single low viscosity master fault of one element width. On the point of interaction the viscosity drops by more than one order of magnitude.
Figure 2. Dual (thermo-elastic and thermo-mechanical) feedback generates extremely narrow low viscosity shear zones (
